While the Pokemon GO 3D digital billboard in Tokyo looks cool, this ‘Text-to-Pokemon’ AI model might just be cooler. Simply put, put in any text prompt and generate your own Pokémon character without “prompt engineering” required. It was trained on BLIP captioned Pokémon images using two NVIDIA RTX A6000 GPUs on the Lambda GPU Cloud for around 15,000 step.
The Stable Diffusion AI model can be used for many things, but it is not perfect at everything, which means getting results in a particular style or appearance often involves a lot of work “prompt engineering”. For example, if you have a particular type of image in mid, you typically have to spend a while crafting an intricate text prompt to actually fine tune the image generation model itself. Try it our here now for unlimited Pokemon character generating fun.
- Vivid 7” OLED screen
- Local co-op, online, and local wireless multiplayer
- 64 GB internal storage (a portion of which is reserved for use by the system)
Fine tuning is the common practice of taking a model which has been trained on a wide and diverse dataset, and then training it a bit more on the dataset you are specifically interested in. This is common practice on deep learning and has been shown to be tremendously effective all manner of models from standard image classification networks to GANs,” said researcher Justin Pinkney.
